A Cholesky factorization based approach for blind FIR channel identification

Abstract

For blind channel identification, various techniques, including transmitter-induced cyclostationarity based approaches, have been proposed in literature. To induce cyclostationarity at the transmitter, zero padding to every symbol packet can be considered. Some subspace methods are proposed for the channel estimation by exploiting the induced cyclostationarity. Due to zero padding, the covariance matrix of received signal vectors can have a special structure. Utilizing this structure, we can estimate channel impulse responses through factorization. In this correspondence, we propose a factorization based approach to estimate channel impulse responses. In general, the proposed factorization based approach can work with a small number of samples and becomes much more computationally efficient than the subspace method when the length of packets is long.